Assessment of Pork Quality Levels by Hyperspectral Imaging Technology

Objective assessment of pork quality is important for meat industry application. Previous studies focused on using color and water content features to classify pork quality levels without considering the texture feature which is one of three factors in the definition of pork quality standards. In this study, a hyperspectral imaging technology which can utilize texture features was presented to develop a high accuracy system for pork quality classification. Texture features were obtained by filtering hyperspectral images with a Gabor filter. Spectral features were extracted from Gabor-filtered images as well as from hyperspectral images directly. The principal component analysis (PCA) was used to compress spectral features over the entire wavelengths (400-1000 nm) into 5 and 10 principal components (PCs). Both K-means clustering and linear discriminant analysis (LDA) were applied to classify pork samples. Results showed that, the accuracy of K-means clustering reached 83% for both 5 combined PCs and 10 combined PCs, which were 15% and 18% higher, respectively, compared to that without using texture features. In order to remove the bias of training set selection in LDA, a total of 210 combinations of training sets were used to obtain the statistical classification results. The average accuracy of LDA reached to 89% for combined 5 PCs, which had a statistically significant improvement comparing to the average accuracy of 73% without using texture features.